Google launches SynthID portal to flag AI-generated content

The system targets content generated by Google’s Gemini, Imagen, Lyria, and Veo models. According to Pushmeet Kohli of Google DeepMind, the detector functions by analyzing uploaded files for hidden markers embedded during the creation process. For audio and visual media, the software highlights precise sections where these watermarks appear, providing a granular look at how the AI contributed to the final output.

Google has restricted initial access to a select group of early testers. A broader rollout is planned for those who join the company's waitlist, with the goal of gathering feedback from professional users to refine transparency standards. While the tool represents a significant step toward labeling machine-made media, its real-world efficacy remains unproven until the public testing phase yields wider results.